为了改善光纤通信系统故障诊断效果, 设计了一种基于模式识别的光纤通信系统故障诊断模型。首先提取光纤通信系统故障诊断的特征向量, 然后采用模式识别技术中的最小二乘支持向量机构建光纤通信系统故障诊断分类器, 最后采用具体光纤通信系统故障诊断实例对模型性能进行测试。结果表明, 设计模型的光纤通信系统故障诊断率超过90%, 诊断效率较传统系统改善明显, 且故障误诊率明显低于对比模型。

采用信息融合的方法, 对高功率微波打击后的装备进行故障诊断。使用 D-S证据理论, 将信息融合应用于高功率微波 (HPM)效应的故障诊断中, 并用实例进行了仿真分析。结果表明: 采用信息融合方法来诊断高功率微波效应后的故障是有效的, 随着融合次数的增加, 故障诊断的确定性增大, 该方法提高了故障诊断的正确率。

In this paper, a hybrid electronic noses’ system (HENS) based on MOS-SAW detection units intended for lung cancer diagnosis is proposed. The MOS gas sensors are used to detect the VOC molecules with low molecular weight (LMW), and the SAW sensors are adopted for the detection of VOC with high molecular weight (HMW). Thus, the novel combination of these two kinds of gas sensors provides higher sensitivities to more of VOC species in breath than that of using only a single kind of sensor. The signals from MOS-SAW detection units are then recognized by a multi-model diagnosis method. Applying four algorithms, six models were established for diagnosis and tested by leave-one-out cross-validation method. The model by artificial neural network (ANN) was selected as the best model to analyze breath samples. 89 clinical samples were tested with MOS-SAW ANN diagnostic model, which takes the features derived from both the MOS and SAW sensors. It shows the highest sensitivity of 93.62%, and the highest selectivity of 83.37%. The study shows that, promisingly, our HENS is effective during screening of lung cancer patients, especially among the people of high risk.